Most health care dollars are spent in the last few months of life, reflecting the compression of illness that occurs. Recent trends towards increased patient autonomy have resulted in the passage of the Patient Self-Determination Act of 1990 where hospitals, nursing homes, and other institutions must, upon admission, inform patients of their rights to (1) declare their advance directives for health care (ADHC) and (2) appoint a health care representative (HCR). However, it is more appropriate for these discussions to take place in the ambulatory environment between the patient and his/her primary care physician before critical illnesses occur. We propose to influence the most personal aspect of medical care, the doctor-patient relationship, with one of the most technologic aspects of medical care, the computerized medical record. We propose to use a state-of-the-art computer-based surveillance/reminder system to encourage physicians to initiate discussions about ADHCs and HCRs, and place signed ADHC and HCR forms in patients' computerized records. We will study the ability of such a reminder system to generate these discussions and subsequent completed forms in a randomized, controlled trial in a large urban academic primary care internal medicine practice. All physicians will be included in the trial as will patients 75 years old or older and patients 55-74 with one or more of selected clinical conditions associated with morbidity and mortality (e.g. cancer, stroke, etc.). The following outcomes will be measured: (1) the number of discussions about ADHCs and/or HCRs that are initiated, (2) the number of ADHC and/or HCR forms that are completed, and (3) patients' satisfaction with their primary care. In the second part of this project, we will assess the effect of electronic ADHC and HCR forms on subsequent care delivered in the emergency room (ER) and in the hospital. The forms will be available through data terminals located throughout the hospital, the ER, and all affiliated clinics and doctors' offices. For patients with completed forms visiting the ER (or hospitalized), we will randomly select patients without forms who visit the ER (or are hospitalized) and compare (1) clinical decisions (e.g. admission to the hospital, ICU admissions, resuscitation, IV therapy, nursing home placement, etc.), (2) 30-day mortality, and (3) ER and inpatient costs. It is the goal of this project to increase the number of patients whose wishes concerning their future health care are documented, and by making these wishes widely available, enable patients to guide their clinical care. If this can be demonstrated, then computer-based medical records will be better able to perform their primary task: improving patient care.